Compositional Data Analysis

Use this page as a guidance-only introduction to compositional constraints and why simple unconstrained analysis can be misleading.

What this method is

This page is a guidance-only introduction to data where components are constrained relative to each other.

Compositional data are not just "small numbers" or "percentages." They are values that live under a shared constraint, often because components stay within a bounded scale and their row-wise totals stay near a fixed total such as 1 or 100.

That constraint matters because changing one component can mechanically change the interpretation of the others. This is why simple unconstrained comparisons can become misleading.

When to use or avoid

Use this page when:

  • your variables represent parts of a whole
  • component values stay within a bounded scale such as 0 to 1 or 0 to 100
  • row-wise totals tend to stay near a common total
  • abundance, fraction, share, or composition language is central to the dataset

Avoid reading this page as proof that the current product already exposes a full compositional analysis pipeline.

  • Do not assume clr, ilr, Dirichlet, or adjacent dedicated methods are available now.
  • Do not treat this page as a substitute for reviewing the current warning and export behavior.

Required inputs

The strongest current implementation evidence is not one dedicated method API. Instead, the repo confirms compositional-risk detection from signals such as:

  • bounded-value ranges
  • near-constant row-wise totals
  • column names that imply fractions, percentages, or abundance
  • request wording that suggests compositional interpretation

That means the current public framing is strongest as method guidance plus warning context.

Outputs

No dedicated compositional-analysis output contract was confirmed in this pass.

The clearest current outputs are adjacent warning and admissibility behaviors, not one canonical inferential result bundle.

No dedicated compositional-analysis figure family was confirmed in this pass.

Interpretation notes

  • Compositional constraints can make simple component-wise comparisons harder to interpret than they first appear.
  • The current repo is stronger on compositional warning and export review than on dedicated compositional modeling.
  • This page should stay narrower than its title might suggest until a true compositional method branch is confirmed.

Limits

  • No dedicated CLR, ILR, Dirichlet, or adjacent compositional-analysis route was confirmed in this pass.
  • This page does not describe a multi-omics workflow or a domain-specific end-to-end pipeline.
  • Stronger claims should wait until method execution support is documented explicitly.

TODO (Phase02+)

  • Confirm dedicated compositional method support before expanding this page beyond guidance-only wording.